| name | spec-audit |
| description | Read-only inventory and sanity audit of the behavioral specs defined in this repo (entities, rules, invariants, scenarios). Runs in an isolated forked context so it doesn't fill the main conversation. Use when asked to review, inventory, or sanity-check the spec definitions. |
| context | fork |
| agent | Explore |
| allowed-tools | Read, Grep, Glob |
Produce a read-only inventory and sanity audit of this project's behavioral
specifications. Do not modify any file.
- Find spec definitions across
src/spec/, micro-specs/, and examples/ by
grepping for the def* forms: (defentity, (defconfig, (defrule,
(definvariant, (defscenario, (defvalueset, (defgenerator,
(defreq.
- Inventory what you find, grouped by file: entities, config, rules,
invariants, scenarios, value sets.
- Flag likely gaps (static, best-effort — no execution):
- rules whose
:sets / :ensures reference a field not present on any entity;
- entity-level invariants that reference
has-many accessors (these need
scenario-level testing);
- entities with a
(member ...) state field but no rule that transitions out
of one of its states (possible dead-end).
- Return a concise summary: counts per category, the per-file inventory, and a
short "potential gaps" list. End with one line on what a deeper check would
add (running
(validate-specs) / (run-pbt) via eval_lisp).
Keep the output compact — this is a survey, not a fix.
